X-Git-Url: https://git.ao2.it/tweeper.git/blobdiff_plain/1e7a09621f810d54f9badc8bcf3f01a6c6ae2347..fae2bb79789ff07d7cdeb6b609b3fd9b43b107b8:/src/Tweeper.php?ds=sidebyside

diff --git a/src/Tweeper.php b/src/Tweeper.php
index e98623b..10e07e3 100644
--- a/src/Tweeper.php
+++ b/src/Tweeper.php
@@ -6,7 +6,7 @@ namespace Tweeper;
  * @file
  * Tweeper - a Twitter to RSS web scraper.
  *
- * Copyright (C) 2013-2018  Antonio Ospite <ao2@ao2.it>
+ * Copyright (C) 2013-2020  Antonio Ospite <ao2@ao2.it>
  *
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -36,7 +36,7 @@ date_default_timezone_set('UTC');
  */
 class Tweeper {
 
-  private static $userAgent = "Mozilla/5.0";
+  private static $userAgent = "Mozilla/5.0 (Linux; U; Android 4.3; en-us; SM-N900T Build/JSS15J)";
   private static $maxConnectionTimeout = 5;
   private static $maxConnectionRetries = 5;
 
@@ -87,6 +87,46 @@ class Tweeper {
   }
 
   /**
+   * Convert Twitter mobile date to the date format expected in a RSS document.
+   */
+  public static function twitterToRssDate($date) {
+    // Twitter uses relative timestamps in minutes for recent tweets.
+    if (preg_match('/^(\d+)m$/', $date, $matches)) {
+      $timestamp = strtotime("+" . $matches[1] . " min", time());
+      if (FALSE === $timestamp) {
+        $timestamp = 0;
+      }
+    }
+    else {
+      /*
+       * In case the time is specified put it after the date,
+       * to make it recognized by strptime().
+       */
+      if (preg_match('/(.*) - (.*)/', $date, $matches)) {
+        $date = $matches[2] . " " . $matches[1];
+      }
+
+      $timestamp = strtotime($date);
+      if (FALSE === $timestamp) {
+        $timestamp = 0;
+      }
+
+      /*
+       * The twitter mobile UI usually only specifies the month and the day, so
+       * strtotime($date) may interpret the date as being in the future.
+       *
+       * If the date is in the future it is probably in the same day but in the
+       * previous year.
+       */
+      if ($timestamp > time()) {
+        $timestamp = strtotime('-1 years', $timestamp);
+      }
+    }
+
+    return Tweeper::epochToRssDate($timestamp);
+  }
+
+  /**
    * Convert string to UpperCamelCase.
    */
   public static function toUpperCamelCase($str, $delim = ' ') {
@@ -119,16 +159,16 @@ class Tweeper {
    */
   private static function getUrlContents($url) {
     $ch = curl_init($url);
-    curl_setopt_array($ch, array(
+    curl_setopt_array($ch, [
       CURLOPT_HEADER => FALSE,
       CURLOPT_CONNECTTIMEOUT => Tweeper::$maxConnectionTimeout,
       // Follow http redirects to get the real URL.
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_COOKIEFILE => "",
       CURLOPT_RETURNTRANSFER => TRUE,
-      CURLOPT_HTTPHEADER => array('Accept-language: en'),
+      CURLOPT_HTTPHEADER => ['Accept-language: en'],
       CURLOPT_USERAGENT => Tweeper::$userAgent,
-    ));
+    ]);
     $contents = Tweeper::curlExec($ch);
     curl_close($ch);
 
@@ -140,7 +180,7 @@ class Tweeper {
    */
   private static function getUrlInfo($url) {
     $ch = curl_init($url);
-    curl_setopt_array($ch, array(
+    curl_setopt_array($ch, [
       CURLOPT_HEADER => TRUE,
       CURLOPT_NOBODY => TRUE,
       CURLOPT_CONNECTTIMEOUT => Tweeper::$maxConnectionTimeout,
@@ -148,7 +188,7 @@ class Tweeper {
       CURLOPT_FOLLOWLOCATION => TRUE,
       CURLOPT_RETURNTRANSFER => TRUE,
       CURLOPT_USERAGENT => Tweeper::$userAgent,
-    ));
+    ]);
 
     $ret = Tweeper::curlExec($ch);
     if (FALSE === $ret) {
@@ -169,7 +209,7 @@ class Tweeper {
    * Generate an RSS <enclosure/> element.
    */
   public static function generateEnclosure($url) {
-    $supported_content_types = array(
+    $supported_content_types = [
       "application/octet-stream",
       "application/ogg",
       "application/pdf",
@@ -188,7 +228,7 @@ class Tweeper {
       "video/mp4",
       "video/mpeg",
       "video/ogg",
-    );
+    ];
 
     $url_info = Tweeper::getUrlInfo($url);
     if (FALSE === $url_info) {
@@ -270,13 +310,13 @@ class Tweeper {
 
     $encoder = new XmlEncoder();
     $normalizer = new ObjectNormalizer();
-    $serializer = new Serializer(array($normalizer), array($encoder));
+    $serializer = new Serializer([$normalizer], [$encoder]);
 
-    $serializer_options = array(
+    $serializer_options = [
       'xml_encoding' => "UTF-8",
       'xml_format_output' => TRUE,
       'xml_root_node_name' => $root_node_name,
-    );
+    ];
 
     $xml_data = $serializer->serialize($data, 'xml', $serializer_options);
     if (!$xml_data) {
@@ -335,7 +375,7 @@ class Tweeper {
     // If there is a host-specific method to get the XML data, use it!
     $get_xml_host_method = 'getXml' . Tweeper::toUpperCamelCase($host, '.');
     if (method_exists($this, $get_xml_host_method)) {
-      $xml_data = call_user_func_array(array($this, $get_xml_host_method), array($html));
+      $xml_data = call_user_func_array([$this, $get_xml_host_method], [$html]);
       $xmlDoc->loadXML($xml_data);
     }
     else {
@@ -394,7 +434,7 @@ class Tweeper {
 
     if (TRUE === $validate_scheme) {
       $scheme = $url["scheme"];
-      if (!in_array($scheme, array("http", "https"))) {
+      if (!in_array($scheme, ["http", "https"])) {
         trigger_error("unsupported scheme: $scheme", E_USER_WARNING);
         return NULL;
       }
@@ -423,7 +463,7 @@ class Tweeper {
 
     $preprocess_html_host_method = 'preprocessHtml' . Tweeper::toUpperCamelCase($host, '.');
     if (method_exists($this, $preprocess_html_host_method)) {
-      $html = call_user_func_array(array($this, $preprocess_html_host_method), array($html));
+      $html = call_user_func_array([$this, $preprocess_html_host_method], [$html]);
     }
 
     $xmlDoc = $this->htmlToXml($html, $host);